- CRI now supports a "runtime_handler" field for RunPodSandboxRequest, used...

  // APIGroup is the group for the resource being referenced.
  // If APIGroup is not specified, the specified Kind must be in the core API group.
  // For any other third-party types, APIGroup is required.
  // +optional
  optional string apiGroup = 1;

  // Kind is the type of resource being referenced
  optional string kind = 2;

  // Name is the name of resource being referenced
  optional string name = 3;
}

### Bug or Regression

- Fix endpoint reconciler not being able to delete the apiserver lease on shutdown ([#114153](https://github.com/kubernetes/kubernetes/pull/114153), [@aojea](https://github.com/aojea)) [SIG API Machinery]

- Fixes an issue that can result in namespaced custom resources being orphaned when their namespace is deleted, if the CRD defining the custom resource is removed concurrently with namespaces being deleted, then recreated. ([#93790](https://github.com/kubernetes/kubernetes/pull/93790), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Apps]

- Fixes an issue that can result in namespaced custom resources being orphaned when their namespace is deleted, if the CRD defining the custom resource is removed concurrently with namespaces being deleted, then recreated. ([#93790](https://github.com/kubernetes/kubernetes/pull/93790), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Apps]

from a file. ([#59666](https://github.com/kubernetes/kubernetes/pull/59666), [@mtaufen](https://github.com/mtaufen))

* [GCP kube-up.sh] Variables that were part of kube-env that were only used for kubelet flags are no longer being set, and are being replaced by the more portable mechanism of the kubelet configuration file. The individual variables in the kube-env metadata entry were never meant to be a stable interface and this release note only applies if you are depending on them....

    This change may require changes to clients that gate pod creation and/or
    scheduling on the node condition `type` `Ready` `status` being `True` for pods
    that need to run prior to the network being ready.
* **Enhance Kubelet QoS**:
  * Pods are placed under a new cgroup hierarchy by default. This feature requires
    draining and restarting the node as part of upgrades. To opt-out set

- Fixes an issue that can result in namespaced custom resources being orphaned when their namespace is deleted, if the CRD defining the custom resource is removed concurrently with namespaces being deleted, then recreated. ([#93790](https://github.com/kubernetes/kubernetes/pull/93790), [@liggitt](https://github.com/liggitt)) [SIG API Machinery and Apps]

- For GCE cluster provider, fix bug of not being able to create internal type load balancer for clusters with more than 1000 nodes in a single zone. ([#89902](https://github.com/kubernetes/kubernetes/pull/89902), [@wojtek-t](https://github.com/wojtek-t)) [SIG Cloud Provider, Network and Scalability]...

container processes if an out-of-memory event occurs. This implementation has shortcomings: for `Guaranteed` pods, memory can not be fully reserved, and the page cache is at risk of being recycled. For `Burstable` pods, overcommitting memory (setting `request` less than `limit` ) could increase the risk of a container being killed when the Linux kernel detects an out of memory condition.

As an alpha feature, Kubernetes v1.22 can use the cgroups v2 API to control memory allocation and isolation....